Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,104 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
language:
|
4 |
+
- en
|
5 |
+
- de
|
6 |
+
library_name: transformers
|
7 |
+
pipeline_tag: text-generation
|
8 |
+
tags:
|
9 |
+
- finetune
|
10 |
+
- sft
|
11 |
+
- dpo
|
12 |
+
- laser
|
13 |
+
- augmentation
|
14 |
+
- german
|
15 |
+
- english
|
16 |
+
- moe
|
17 |
+
---
|
18 |
+
![SauerkrautLM](https://vago-solutions.de/wp-content/uploads/2024/02/Sauerkraut_Laserchat-14b-MoE.png "SauerkrautLM-14b-MoE-LaserChat")
|
19 |
+
## VAGO solutions SauerkrautLM-14b-MoE-LaserChat
|
20 |
+
Introducing **SauerkrautLM-14b-MoE-LaserChat** – our Sauerkraut (2x7b) 14b MoE version of the powerful [SauerkrautLM-7b-LaserChat](https://huggingface.co/VAGOsolutions/SauerkrautLM-7b-LaserChat) and [yam-peleg/Experiment26-7B](https://huggingface.co/yam-peleg/Experiment26-7B) !
|
21 |
+
|
22 |
+
By combining the two models, we were able to significantly increase both the German and English language skills.
|
23 |
+
In addition, the initial SauerkrautLM-7b-LaserChat also acts as an adapter for Experiment26-7B, which means it benefits from the chat capabilities of the SauerkrautLM-7b-LaserChat.
|
24 |
+
At the same time, the SauerkrautLM-7b-LaserChat benefits from the knowledge and creativity of Experiment26-7B.
|
25 |
+
|
26 |
+
The model **SauerkrautLM-14b-MoE-LaserChat** is a **joint effort** between **VAGO solutions** and **Hyperspace.ai.**
|
27 |
+
Much appreciation goes to the tremendous research effort of **Fernando Fernandes Neto, David Golchinfar and Eric Hartford on their laserRMT approach.**
|
28 |
+
Without their independent research collaboration this model release would not have been possible.
|
29 |
+
|
30 |
+
|
31 |
+
# Table of Contents
|
32 |
+
1. [Overview of all SauerkrautLM-14b-MoE-LaserChat models](#all-sauerkrautlm-14b-MoE-laserchat-models)
|
33 |
+
2. [Model Details](#model-details)
|
34 |
+
- [Prompt template](#prompt-template)
|
35 |
+
- [Training procedure](#proceed-of-the-training)
|
36 |
+
3. [Evaluation](#evaluation)
|
37 |
+
5. [Disclaimer](#disclaimer)
|
38 |
+
6. [Contact](#contact)
|
39 |
+
7. [Collaborations](#collaborations)
|
40 |
+
8. [Acknowledgement](#acknowledgement)
|
41 |
+
|
42 |
+
|
43 |
+
## All SauerkrautLM-14b-MoE-LaserChat Models
|
44 |
+
|
45 |
+
| Model | HF | GPTQ | GGUF | AWQ |
|
46 |
+
|-------|-------|-------|-------|-------|
|
47 |
+
| SauerkrautLM-14b-MoE-LaserChat | [Link](https://huggingface.co/VAGOsolutions/SauerkrautLM-14b-MoE-LaserChat) | coming soon | coming soon | coming soon |
|
48 |
+
|
49 |
+
## Model Details
|
50 |
+
**SauerkrautLM-14b-MoE-LaserChat**
|
51 |
+
- **Model Type:** SauerkrautLM-14b-MoE-LaserChat is a MoE Model based on [SauerkrautLM-7b-LaserChat](https://huggingface.co/VAGOsolutions/SauerkrautLM-7b-LaserChat) and [yam-peleg/Experiment26-7B](https://huggingface.co/yam-peleg/Experiment26-7B)
|
52 |
+
- **Language(s):** German, English
|
53 |
+
- **License:** Apache 2.0
|
54 |
+
- **Contact:** [VAGO solutions](https://vago-solutions.ai), [Hyperspace.computer](https://hyperspace.computer/)
|
55 |
+
|
56 |
+
|
57 |
+
We improved the German language skills on this model further. Nevertheless, certain formulations may occur that are not entirely correct.
|
58 |
+
|
59 |
+
|
60 |
+
### Prompt Template:
|
61 |
+
```
|
62 |
+
GPT4 Correct User: Hallo, wie geht es dir?<|end_of_turn|>GPT4 Correct Assistant: Hallo! Ich bin ein künstliches Intelligenzsystem und habe keine persönlichen Gefühle oder körperliche Zustände. Wie kann ich Ihnen helfen?<|end_of_turn|>GPT4 Correct User: Ich benötige nur einen kurzen Satz, den ich in das Prompt Template veröffentlichen kann.<|end_of_turn|>GPT4 Correct Assistant:
|
63 |
+
|
64 |
+
|
65 |
+
```
|
66 |
+
|
67 |
+
|
68 |
+
```
|
69 |
+
GPT4 Correct User: Hello<|end_of_turn|>GPT4 Correct Assistant: Hello! How can I help you today? If you have any questions or need assistance, feel free to ask.<|end_of_turn|>GPT4 Correct User: I just need a short sentence to post in the prompt template.<|end_of_turn|>GPT4 Correct Assistant:
|
70 |
+
|
71 |
+
```
|
72 |
+
|
73 |
+
|
74 |
+
## Evaluation
|
75 |
+
|
76 |
+
**Open LLM Leaderboard:**
|
77 |
+
|
78 |
+
benchmarked on lm-evaluation-harness 0.4.1
|
79 |
+
|
80 |
+
| Metric | Value |
|
81 |
+
|-----------------------|---------------------------|
|
82 |
+
| Avg. | 71.65 |
|
83 |
+
| ARC (25-shot) | 68.09 |
|
84 |
+
| HellaSwag (10-shot) | 84.78 |
|
85 |
+
| MMLU (5-shot) | 63.59|
|
86 |
+
| TruthfulQA (0-shot) | 58.57 |
|
87 |
+
| Winogrande (5-shot) | 80.74 |
|
88 |
+
| GSM8K (5-shot) | 74.15 |
|
89 |
+
|
90 |
+
|
91 |
+
|
92 |
+
## Disclaimer
|
93 |
+
We must inform users that despite our best efforts in data cleansing, the possibility of uncensored content slipping through cannot be entirely ruled out.
|
94 |
+
However, we cannot guarantee consistently appropriate behavior. Therefore, if you encounter any issues or come across inappropriate content, we kindly request that you inform us through the contact information provided.
|
95 |
+
Additionally, it is essential to understand that the licensing of these models does not constitute legal advice. We are not held responsible for the actions of third parties who utilize our models.
|
96 |
+
|
97 |
+
## Contact
|
98 |
+
If you are interested in customized LLMs for business applications, please get in contact with us via our websites. We are also grateful for your feedback and suggestions.
|
99 |
+
|
100 |
+
## Collaborations
|
101 |
+
We are also keenly seeking support and investment for our startups, VAGO solutions and Hyperspace where we continuously advance the development of robust language models designed to address a diverse range of purposes and requirements. If the prospect of collaboratively navigating future challenges excites you, we warmly invite you to reach out to us at [VAGO solutions](https://vago-solutions.de/#Kontakt), [Hyperspace.computer](https://hyperspace.computer/)
|
102 |
+
|
103 |
+
## Acknowledgement
|
104 |
+
Many thanks to [yam-peleg](https://huggingface.co/yam-peleg) for providing such valuable model to the Open-Source community
|